WERM

WERM is a 302 neuron network built on the real wiring diagram of C. elegans steers a local language model in the browser.

WERM runs a 302 neuron network based on the real wiring diagram of C. elegans inside your browser or on your own computer. The recurrent network modulates a local language model by updating dials like temperature, top_p, and length limits based on the worm's live state. It features a genome drawn to scale, interactive visualisations of the connectome, and body simulations. You can run it completely offline using Node 20 or newer, connecting it to local engines such as Ollama or LM Studio. The software uses real data from published research, including electron microscope wiring diagrams and NCBI reference sequences.

WERM at a glance

Best for: Software engineer, Researcher
Runs on: Web, Node.js, CLI, self-hosted, open source
Pricing: Free
Released: October 2026
Category: AI research tool
Website: werm.si

Key features

302 neuron network based on C. elegans wiring
Interactive 3D globe of neuron connections
Liquid time constant model neuron gates
Local language model steering layer
Genome drawn to scale from NCBI references
Offline browser and Node.js execution
Integration with Ollama and LM Studio

Use it for

Steer local language models with worm brain states
Explore the C. elegans connectome and neuron firing
Run neural simulations and tests locally
Inspect scaled chromosomes and gene annotations

Alternatives to WERM

Neuron Explorer — It keeps every rollout and maps model layers in perspective.
Next Brain — Executes neural models across CPU, GPU, browser and MPI runtimes.
WeKnora — Combines RAG, multi-step agents, and automatic Wiki generation in one platform.
NanoJev — It runs parallel decisions and outputs probability distributions without output-token decoding.
Ncm — Generates text significantly faster than diffusion baselines while maintaining high quality.
Every number in a tiny GPT — Inspects every lookup, matrix multiplication, and normalisation in a live transformer.
Biomni — Automates complex wet-lab experimental design with specialized tools and databases.
Auto Deep Researcher — Automates experiment operations to reduce manual babysitting of training loops.
Arbor — It grows a hypothesis tree and tests changes on held-out data.
Overview — Learn language models by running hands-on experiments without writing scripts
Vegaml — Runs locally on your hardware with calibrated probabilities and conformal answer sets.
Hypoarena — Provides a fully offline environment to test scientific discovery pipeline logic.

All WERM alternatives